From Fragmentation to Integration: A Delphi-Validated AI Literacy Competency Framework for Higher Education Teachers in China

Higher education is currently navigating a profound shift. Generative artificial intelligence, the technology behind tools that can write essays, solve complex problems, and hold conversations, has arrived in universities with startling speed. For teachers, this technology presents a dual reality: it offers powerful ways to personalize learning and support research, but it also introduces serious questions about academic honesty, data privacy, and the very nature of learning. To navigate this new landscape, educators need more than just the ability to press buttons; they need a deep understanding of how these tools work, how to use them ethically, and how to integrate them into their teaching without losing their own professional agency. While governments and universities have begun to draft guidelines for this new era, these instructions often exist in isolation, creating a patchwork of rules that can be confusing and inconsistent.

In response to this fragmentation, a team of researchers set out to build a unified map for Chinese university teachers. They sought to create a single, coherent framework that would define what it means to be literate in artificial intelligence within the specific context of Chinese higher education. The researchers did not simply guess what teachers needed; they began by carefully comparing four existing official documents and policy frameworks from China, alongside two international models. This initial analysis revealed a significant disconnect. Some documents treated AI as a simple tool to be used, while others viewed it as a partner in the learning process. Some listed skills as a static checklist, while others tried to describe a path of growth. The researchers identified three main tensions holding the field back: the struggle between creating one standard rule for everyone versus allowing for local context, the conflict between listing skills as a fixed set versus showing how they develop over time, and the disagreement over whether AI is merely a machine or an active participant in education.

To resolve these tensions, the researchers synthesized a preliminary framework consisting of twenty-seven specific skills and understandings, organized into three distinct layers of development. The first layer, Foundation, covers the basics: understanding what AI is, knowing how to use the tools, and grasping the ethical rules. The second layer, Practice, focuses on the classroom: designing lessons that use AI, creating new learning materials, and assessing students fairly in an AI-rich environment. The third layer, Innovation, looks toward the future: redesigning entire courses, leading research, and mentoring colleagues. Crucially, the researchers placed ethics not as a single box to check, but as a thread running through all three layers, ensuring that responsible use is considered at every step.

This draft framework was then put to the test through a rigorous process of expert consultation known as a Delphi study. The researchers invited fifteen leading scholars and university administrators, all with deep experience in education and technology, to review the proposed skills. In the first round, these experts rated each of the twenty-seven indicators for importance and clarity. The feedback was immediate and specific. The experts suggested merging some overlapping items, removing one that was redundant, and adding three new, critical skills that had been missing, such as using AI to make teaching more accessible for students with disabilities. They also insisted that the ethical dimension be elevated to a cross-cutting principle rather than just another item on a list.

In the second round, the experts reviewed the revised framework. The results were decisive. Twenty-six of the twenty-seven indicators achieved a strong consensus, meaning the experts agreed they were essential and clearly defined. These twenty-six skills now form the core of a validated competency framework for Chinese university teachers. However, one indicator regarding the management of classroom dynamics when students use AI did not reach the required level of agreement. The experts felt this area was still too new and uncertain to be declared a standard requirement. Consequently, the researchers kept this specific skill separate, labeling it as a provisional item that requires further study before it can be formally adopted.

The final result is a structured, evidence-based guide that moves beyond a simple list of technical skills. It offers a developmental path, acknowledging that teachers start with basic awareness and can grow into leaders who redesign their institutions' approach to technology. The framework extends existing international models by explicitly tailoring them to the Chinese context and by organizing competencies into a clear progression from learning the basics to innovating with the technology. While this framework provides a solid foundation for universities to design training programs and for policymakers to set standards, the researchers caution that it is not yet a finished measurement tool. It has been validated by expert opinion, but it has not yet been tested in the classroom to see if it accurately predicts how well a teacher will perform. The next step for the field is to use this framework to build actual tests and observe teachers in action, ensuring that the map they have drawn truly leads to the destination of effective, ethical, and human-centered teaching in the age of artificial intelligence.